Since the author seems interested in the maximum number of moves required to solve the puzzle, a similar puzzle called Subway Shuffle far outdoes Rush Hour. For example, puzzle 100 involves 9 pieces on a 10-spot grid, but requires (as far as is known, maybe the solution isn't optimal?) 589 steps to solve. https://www.cs.brandeis.edu/~storer/JimPuzzles/ZPAGES/zzzSub...
I had the app installed way back when, and it was fun. I reached my limit around 60-90 step solutions. The numbers are a little deceiving: in many cases there are sequences of moves 3, 4, or 5 (or more?) long where only one move is possible, and therefore in a sense they qualify as only one decision as a unit, even if they are multiple moves. I think this same aspect is sometimes true of Rush Hour, but less so.
As far as I know it was iOS-only. I think the author created the puzzle as a phd paper, and once I found the paper itself online. It was interesting reading.
It should be possible today to vibe code the mechanics as a web app, right?
The Rush Hour puzzle is quite fun when viewed as a planning problem. In standard PDDL the model becomes very messy. I like the
extensions proposed in https://arxiv.org/abs/2412.06312v1 that makes the model intuitive.
It turns out that Rush Hour becomes much harder if we shrink the cars from size 2x1 to size 1x1, while maintaining their direction to be either horizontal or vertical [1].
While the hardest 6x6 Rush Hour puzzle takes 51 moves, the hardest Unit Rush Hour puzzle takes a whopping 732 moves [2].
9 comments
[ 3.5 ms ] story [ 32.7 ms ] threadIt also seems like something that would be great as part of Simon Tatham's puzzle collection: https://www.chiark.greenend.org.uk/~sgtatham/puzzles/ / https://play.google.com/store/apps/details?id=name.boyle.chr...
As far as I know it was iOS-only. I think the author created the puzzle as a phd paper, and once I found the paper itself online. It was interesting reading.
It should be possible today to vibe code the mechanics as a web app, right?
The Rush Hour puzzle is quite fun when viewed as a planning problem. In standard PDDL the model becomes very messy. I like the extensions proposed in https://arxiv.org/abs/2412.06312v1 that makes the model intuitive.
[1] https://tromp.github.io/orimaze.html
[2] https://tromp.github.io/rh.ps
PS: Good chance that if you're reading these comments that you will appreciate this video by 2swap, visualising solutions to Rush Hour in 3D: https://youtu.be/YGLNyHd2w10?si=fGFqzEbmV3utbA0O
Ive just released a game built around this project on play store - https://play.google.com/store/apps/details?id=com.noobgiraff...
Has 4x4, 5x5 and 6x6 puzzles. Generating 7x7 puzzles is long process and i think i may have to do it on cloud, but am planning to release them soon.